- Rating
- 2
- Confidence
- 2
- Informativeness
- 2
- Presentation
- 2
- Soundness
- 2
Read review
Paper summary
The paper develops a theoretical framework for analyzing the convergence and stability of iterative refinement in DDDMs using residual energy, Jacobian analysis, and strong-convexity conditions. Its main contribution is identifying why existing stability certificates are overly conservative and proposing tighter curvature-based bounds. Experiments across image and molecular diffusion models validate the theory, showing rapid refinement convergence and substantial gaps between theoretical certificates and actual Hessian geometry. The results also reveal how training objectives and diffusion settings influence refinement stability.
Critical strengths
The paper provides a principled analysis of DDDM refinement stability through fixed-point contraction and strong-convexity conditions. This strengthens the theoretical understanding of deterministic diffusion sampling and offers useful tools for analyzing convergence behavior.
Critical weaknesses
- While the paper provides theoretical convergence guarantees and tighter stability certificates, it does not establish a clear connection between these properties and improvements in generation quality, sampling efficiency, or diversity. This limits the practical significance of the proposed framework.
- The proposed certificates remain overly conservative, failing to detect strong convexity in many cases where it is empirically observed. Moreover, the guarantees do not extend to the evaluated molecular diffusion models, raising concerns about the framework's applicability beyond specific image diffusion settings.
Questions for authors
-
Do the proposed stability certificates correlate with reconstruction quality, sampling efficiency, or generation diversity? Evidence of such a relationship would help establish the practical relevance of the theoretical results.
-
Why do the proposed certificates fail for molecular diffusion models? Is this primarily due to overly conservative bounds or fundamentally different refinement dynamics? Clarifying this distinction would help assess the applicability of the theoretical framework.
-
In the experiments, why do you use 3999 sampling steps, which is not common in diffusion sampling?
7 comments
Correlation with molecular reconstruction quality
To further investigate whether the stability measures are related to the physical properties of generated molecular structures, we examined the correlations between the stability-related quantities and molecular geometry. We found that the minimum Hessian eigenvalue, , was positively correlated with mean bond length () and mean and median bond angles (). The gradient-normalized minimum Hessian eigenvalue showed positive correlations with both bond-length and bond-angle measurements (--). We also observed some negative relationships. For example, the Lipschitz constant of the reconstruction map, , was negatively correlated with maximum bond length (), while the Hessian symmetry error was negatively correlated with maximum bond angle (). These observations suggest that some stability-related quantities are associated with variations in molecular reconstruction quality.
Table 1. Strongest positive correlations between stability measures and molecular geometry.
| Stability measure | Geometry metric | Spearman correlation () |
|---|---|---|
| Minimum Hessian eigenvalue () | Mean bond length | +1.0 |
| Gradient-normalized minimum Hessian eigenvalue | Median bond length | +1.0 |
| Gradient-normalized minimum Hessian eigenvalue | Median bond angle | +1.0 |
| Gradient-normalized minimum Hessian eigenvalue | Mean bond angle | +1.0 |
| Minimum Hessian eigenvalue () | Median bond angle | +0.9 |
| Minimum Hessian eigenvalue () | Mean bond angle | +0.9 |
| Gradient-normalized minimum Hessian eigenvalue | Mean bond length | +0.9 |
| Minimum Hessian eigenvalue () | Median bond length | +0.9 |
Table 2. Additional observed correlations.
| Stability measure | Geometry metric | Spearman correlation () |
|---|---|---|
| Contraction margin ()^2 | Mean bond angle | +0.6 |
| Hessian symmetry error | Mean bond length | -0.7 |
| Minimum Hessian eigenvalue () | Maximum bond angle | +0.6 |
| Curvature-related residual | Maximum bond length | -0.7 |
| Contraction margin ()^2 | Median bond length | +0.6 |
| Reconstruction residual () | Maximum bond length | +0.7 |
| Reconstruction residual () | Minimum non-bonded distance | +0.7 |
| Lipschitz constant of the reconstruction map () | Minimum non-bonded distance | -0.6 |
| Reconstruction residual () | Mean bond length | -0.6 |
| Curvature measure | Maximum bond length | -0.7 |
| Lipschitz constant of the reconstruction map () | Median bond length | +0.6 |
Table 3. Molecular geometry measurements for certified molecules ().
| Geometry metric | Certified () |
|---|---|
| Mean bond length | 0.029734 |
| Median bond length | 0.022467 |
| Maximum bond length | 0.110362 |
| Mean bond angle | 3.543294 |
| Median bond angle | 2.960807 |
| Maximum bond angle | 15.940434 |
| Number of clashes | 0 |
| Fraction of clashes | 0 |
| Minimum non-bonded distance | 1.741740 |
The positive correlations indicate that molecules with higher values of certain stability measures also tend to exhibit larger values of the corresponding bond-length and bond-angle measurements. For the three certified molecules, the mean bond-length deviation was 0.0297 Å, and the mean bond-angle deviation was 3.54°, with no detected non-bonded clashes. These measurements provide a descriptive characterization of the reconstructed geometries.
However, given the small number of certified molecules, we do not interpret the observed correlations between stability measures and molecular geometry as statistically meaningful. In particular, these results are insufficient to establish a general relationship between the proposed stability conditions and physical reconstruction quality. In particular, these observations are insufficient to establish a general relationship between the proposed stability conditions and physical reconstruction quality. A larger evaluation would be needed to investigate such relationships reliably.
Molecule Refinement Dynamics
We traced the molecular refinement instability to the way the refinement head is implemented. For each node, the network predicts a single scalar correction, which the sampler broadcasts across all 74 node features. Thus, the refinement map can be written as , where is the scalar-valued refinement head and is fixed during refinement. Differentiating this map gives . Since this is a rank-one outer product, its spectral norm is exactly . Consequently, contraction requires . In our experiments, the measured gradient norms range from 0.145 to 7.10, exceeding this threshold in all 160 evaluated configurations.
We also evaluated a model with a 74-dimensional refinement head to check whether predicting separate corrections for each feature would improve stability. This removes the amplification introduced by scalar broadcasting: the ratio is 1.0 across all 160 configurations, compared with approximately 8.6 for the scalar head. However, the refinement remains unstable: the measured Jacobian norms range from 1.01 to 10.74, and none of the evaluated states satisfies the contraction condition. Since the two models also differ in their encoders, training objectives, and budgets, this is not a controlled ablation. Nevertheless, it shows that removing the scalar-broadcasting amplification alone is not sufficient to ensure stable refinement.
We identified an additional feedback pathway in the molecular sampler: the running estimate enters the network both as the refinement input and through the node features, which are overwritten with the broadcast estimate after each refinement step. Consequently, the Jacobian governing the refinement map in Proposition 2 captures only the first pathway, while the implemented sampler also propagates the estimate through its conditioning inputs.
We empirically isolated these two pathways across five model checkpoints. At epoch 20, for example, the estimate-input Jacobian norm,, is numerically zero, and refinement converges in two steps when the node features are held fixed. However, the full sampler still diverges. This demonstrates that contraction with respect to the explicit refinement input alone is insufficient to guarantee stability when the estimate also re-enters through the conditioning features.
We therefore clarify this as a scope condition of Proposition 2 rather than a counterexample to the proposition: its contraction guarantee applies to the refinement map under the stated input dependence, but does not account for the additional conditioning feedback introduced by this molecular sampler. Unlike the molecular implementation, the image-model samplers considered in our experiments do not introduce this additional feedback pathway.
For molecular diffusion on GEOM-LongRange (molecules with more than 100 atoms), the sufficient condition in Proposition 2 is not satisfied across the evaluated diffusion states: the clamped certificate numerator Δ=max(0,1−L)2−RBsq is non-positive, resulting in μbound≤0. This means that our bound cannot certify local strong convexity for the evaluated molecular models, although it does not imply that their reconstruction energies are necessarily non-convex. To investigate the role of the curvature estimate, we compare the original B_sq-based estimate with the direction-aware quantity R * ||H_u||_op. Relative to MSE, Pseudo-LPIPS reduces the estimate-slot Lipschitz constant by 78% and 39%, and the residual-weighted directional curvature by 87% and 40%, under the VP-cosine and VE schedules, respectively. Notably, under VP-cosine, the B_sq-based certificate remains unchanged despite the substantial reduction in directional curvature, whereas the direction-aware certificate captures this change.
The generic curvature bound may be too conservative to capture changes in the local geometry of molecular reconstruction. However, since the sufficient condition remains unsatisfied, we cannot determine whether its failure is entirely due to bound conservativeness or also reflects the local geometry of the molecular models. Thus, a non-positive certificate indicates that strong convexity cannot be guaranteed, not that the reconstruction energy is necessarily non-convex.